Cascade interference

_______________________________________________________________________________Discussion

83 As a consequence, it is plausible that Cas6, Cas5 and Cas7 can be unspecifically complexed with E. coli RNA. Thus, individual assembly experiments have to be compared to the wildtype assembly. Defective Cascade formation around modified crRNAs is reflected in a relatively low amount of crRNA uptake. This suggests a low affinity to the respective crRNA which results in the observation of monomeric Cas7 protein and the formation of Cas7 oligomers with E. coli RNA. In addition to this, Cas6 is present in the assembly using the 3'G RNA, whereas it is absent from the assembly that was performed with the 5'G RNA. According to the data discussed in section 4.1, it would be assumed that the modification of the 3'-terminal tag hinders Cas6 interaction with the crRNA. The fact that Cas6 is not associated with the Cas protein assembly using the crRNA with a modified 5'-terminal tag indicates hindered complex formation. In contrast, the assembly experiment using a modified 3'-terminal tag showed proper Cascade formation, which indicates that Cas6 is rather associated on a protein:protein level to Cas7 than via crRNA contact in the in vitro studies. In order to get a better understanding of crRNA:Cas protein interactions, the use of modified crRNAs and subsequent in vitro assembly monitoring seems rather challenging as the crRNA binding Cas proteins interact unspecifically with RNA in vitro.

As the crystallographic structures of the crRNA binding Cas protein homologues are available, it seems plausible to perform a single molecule Förster resonance energy transfer (smFRET) analysis, wherein dye molecules could be incorporated at predicted interaction sites between Cas proteins as well as crRNA, which could be used to monitor the order complex formation. Additionally, this would also give a better understanding on the localization of the crRNA within the Cascade complex ¹⁴⁵.

_______________________________________________________________________________Discussion

84 nuclease domain of Cas3 to nick the target DNA that activates the helicase which is responsible for ATP-dependent dsDNA hydrolysis ^{72, 148}. As described in section 2.4.5, a Cascade-mediated interference reaction could not be observed for the type I-B Cascade from C. thermocellum in vitro. Several factors could be responsible for this.

First, the PAM sequences used in the interference assays could differ from the one that mediates target recognition, as the PAM sequence specific for C. thermocellum has not been determined yet. The fact that the PAM sequences that were identified for the type I-B CRISPR systems of Haloferax volcanii and Listeria monocytogenes differ in sequence, indicates that the PAM is not subtype specific but has to be identified for each organism individually ^{113, 114}. The CRISPRtarget tool was utilized to identify potential DNA targets of the CRISPR system in C. thermocellum. Three potential PAM sequences were identified adjacent to protospacers from different prophages/plasmids of Clostridia species. Out of these, the 3'-AGT-5' PAM sequence represents the most promising hit as it was found in a Clostridium thermocellum prophage. However, one viral evasion strategy to escape CRISPR-Cas is to introduce variations in the PAM sequence in the own genome ⁴⁴. Thus, it is possible that the PAM sequences identified in the prophages/plasmid represents a modified PAM that can escape Cascade interference. Recently, EMSA experiments performed with Cas8 from Methanothermobacter thermautotrophicus showed that the protein responds to a 3'-GGG-5' PAM in dsDNA, which could also be tested for interference assays using recombinant Cascade from C. thermocellum.

Second, R-loop stabilization by the large subunit Cas8b of the complex could be inefficient. It was shown that Cas8b co-purifies together with a protein fragment that represents an additional C-terminal part of Cas8b. Consequently, the question arises whether this protein fragment plays a functional role or if it is artificially produced and hinders interference. To exclude internal translation and proteolytic self-cleavage of the protein, mutations were introduced within the region of a potential RBS adjacent to first amino acids of the Cas8b fragment and N-terminal protein truncations were generated.

Analysis of the respective mutants revealed that production of the protein fragment is not prevented. Western blot analysis using a poly-clonal anti-Cas8b antibody and C.

thermocellum cell extract confirmed the production of the additional protein fragment in vivo and excludes proteolytic cleavage of Cas8b during heterologous expression in E. coli.

Interestingly, the expression of a truncated Cas8b protein, lacking the C-terminal part that

_______________________________________________________________________________Discussion

85 represents the small fragment only yielded insoluble protein. This could indicate that the small fragment is produced by a split of the full-length Cas8b protein which causes subsequent precipitation of the remaining N-terminal part of the protein (Masterthesis of Kristina Rau). To investigate whether Cas8b is a functional large subunit of Cascade in the presence of the small protein fragment, EMSAs could be performed using Cascade and radiolabeled dsDNA targets to ensure target binding. Furthermore, an enzymatic or chemical footprint analysis specific for ssDNA using Cascade and the respective dsDNA target could confirm proper R-loop formation ^{49, 149}. Alternatively, once the crystal structure of the Cas8b protein is solved, smFRET could be used to monitor R-loop formation and investigate the interaction between the different copies of Cas8b within the complex ¹⁴⁵.

A final explanation for the absence of Cascade DNA interference activity is the possibility that the Cas3 protein of C. thermocellum is not active in vitro. Recombinant Cas3 was produced with a 6-fold C-terminal His-tag and purified from E. coli. It is possible that the protein is misfolded due to the addition of the His-tag, but still remains soluble.

Furthermore, the recombinant Cas3 protein could be loaded with unspecific nucleic acids from E. coli. The crystal structure of the Cas3 protein from Thermobifida fusca revealed ssDNA that was captured in the precleavage state. The DNA was of endogenous origin and copurified with the protein ¹⁵⁰. An additional purification step using ion-exchange chromatography could not be applied to Cas3 of C. thermocellum, as the protein started to precipitate shortly after affinity purification. Cas3 contains an N-terminal HD helicase domain which represents the proposed metal-dependent exo- and endonuclease ⁷³. The crystal structures of different Cas3 homologues revealed the interaction of iron (II), nickel or calcium ions with the HD residues 73, 150, 151

. In addition to this, Cas3 ssDNAse activity was observed with divalent magnesium, manganese, cobalt, copper and zinc ions 72, 73, 150, 152. Hence, a wide range of cofactors could possibly be required for the activation of the Cas3 HD nuclease ⁵⁷. The nuclease assay using a ssDNA substrate with increasing concentrations of Cas3, magnesium and manganese ions showed that even at a 4 µM protein concentration, the substrate is not cleaved efficiently. This could indicate that magnesium and manganese are not the corresponding Cas3 cofactors. Therefore, the Cas3 cleavage assay was performed under varying metal conditions. Efficient DNA degradation

_______________________________________________________________________________Discussion

86 could not be observed under any of the used metal conditions. Hence, further metal combinations have to be tested to determine optimal cleavage conditions for Cas3.